211776 2020-05-12 03:44:54 -0700 REGRESSION (iOS 13.4.1): SpeechSynthesisUtterance.onend event won't fire on cancel(). 2020-05-21 10:50:20 -0700 1 1 1 Unclassified WebKit Accessibility Safari 13 All iOS 13 RESOLVED FIXED InRadar P2 Normal --- 1 baba cfleizach ap bfulgham cfleizach pvollan webkit-bug-importer oldest_to_newest 1651675 0 399117 baba 2020-05-12 03:44:54 -0700 Created attachment 399117 testcase Step to reproduce: 1. Open attached test.html 2. Click "play" button. 3. Click "stop" button before the speaking finished. Expected behavior: An alert "on end" should be shown. Actual result: No alert is shown. UA: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_4) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/13.1 Safari/605.1.15 Remarks: This issue occurs on iOS 13.4.1, while iOS 13.3.1 does not have the issue. Able to reproduce on both simulator and real device. (Probably Safari on macOS has same issue, but details have not been confirmed.) 1651676 1 webkit-bug-importer 2020-05-12 03:45:03 -0700 <rdar://problem/63130249> 1651730 2 cfleizach 2020-05-12 08:03:13 -0700 (In reply to Takao Baba from comment #0) > Created attachment 399117 [details] > testcase > > Step to reproduce: > 1. Open attached test.html > 2. Click "play" button. > 3. Click "stop" button before the speaking finished. > > Expected behavior: > An alert "on end" should be shown. > > Actual result: > No alert is shown. > > UA: > Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_4) AppleWebKit/605.1.15 (KHTML, > like Gecko) Version/13.1 Safari/605.1.15 > > Remarks: > This issue occurs on iOS 13.4.1, while iOS 13.3.1 does not have the issue. > Able to reproduce on both simulator and real device. > > (Probably Safari on macOS has same issue, but details have not been > confirmed.) It looks like "onerror" is fired when a cancel event is generated. I think this is correct based on spec usage 4.2.7. SpeechSynthesisErrorEvent Attributes The SpeechSynthesisErrorEvent is the interface used for the SpeechSynthesisUtterance error event. error attribute, of type SpeechSynthesisErrorCode, readonly The errorCode is an enumeration indicating what has gone wrong. The values are: "canceled" A cancel method call caused the SpeechSynthesisUtterance to be removed from the queue before it had begun being spoken. https://wicg.github.io/speech-api/ 1652064 3 399215 baba 2020-05-12 18:09:25 -0700 Created attachment 399215 testcase2 1652065 4 baba 2020-05-12 18:12:18 -0700 OK, I've understood that "onerror" should be fired rather than "onend". But neither onend nor onerror seem be called (see testcase2). 1652143 5 cfleizach 2020-05-13 00:27:22 -0700 (In reply to Takao Baba from comment #4) > OK, I've understood that "onerror" should be fired rather than "onend". > > But neither onend nor onerror seem be called (see testcase2). Yep looks like a regression 1652146 6 399247 cfleizach 2020-05-13 01:01:21 -0700 Created attachment 399247 patch 1653140 7 399247 ap 2020-05-15 12:31:35 -0700 Comment on attachment 399247 patch View in context: https://bugs.webkit.org/attachment.cgi?id=399247&action=review > Source/WebCore/ChangeLog:3 > + REGRESSION (iOS 13.4.1): SpeechSynthesisUtterance.onend event won't fire on cancel(). Obvious question: can a regression test be made for this issue? 1653143 8 399247 cfleizach 2020-05-15 12:33:24 -0700 Comment on attachment 399247 patch View in context: https://bugs.webkit.org/attachment.cgi?id=399247&action=review >> Source/WebCore/ChangeLog:3 >> + REGRESSION (iOS 13.4.1): SpeechSynthesisUtterance.onend event won't fire on cancel(). > > Obvious question: can a regression test be made for this issue? we have one for the Mock synthesizer that goes all through WebCore. This one uses the synthesizer in the client, which AFAICT there were no tests added for. @Per, @Brent, is there a way to test this flow? Right now we use a mock synthesizer in these tests that bypasses this logic that was fixed 1653218 9 399247 bfulgham 2020-05-15 16:05:19 -0700 Comment on attachment 399247 patch View in context: https://bugs.webkit.org/attachment.cgi?id=399247&action=review >>> Source/WebCore/ChangeLog:3 >>> + REGRESSION (iOS 13.4.1): SpeechSynthesisUtterance.onend event won't fire on cancel(). >> >> Obvious question: can a regression test be made for this issue? > > we have one for the Mock synthesizer that goes all through WebCore. This one uses the synthesizer in the client, which AFAICT there were no tests added for. > > @Per, @Brent, is there a way to test this flow? Right now we use a mock synthesizer in these tests that bypasses this logic that was fixed I do not know of a way to test the system speech synthesizer flow. 1653220 10 399247 pvollan 2020-05-15 16:16:22 -0700 Comment on attachment 399247 patch View in context: https://bugs.webkit.org/attachment.cgi?id=399247&action=review R=me. >>> Source/WebCore/ChangeLog:3 >>> + REGRESSION (iOS 13.4.1): SpeechSynthesisUtterance.onend event won't fire on cancel(). >> >> Obvious question: can a regression test be made for this issue? > > we have one for the Mock synthesizer that goes all through WebCore. This one uses the synthesizer in the client, which AFAICT there were no tests added for. > > @Per, @Brent, is there a way to test this flow? Right now we use a mock synthesizer in these tests that bypasses this logic that was fixed I am not sure how to write a test for this. Is it possible to write a layout test that does not use a mock synthesizer? > Source/WebCore/ChangeLog:17 > + (WebCore::SpeechSynthesis::cancel): Perhaps you could include an explanation here if creating a test is not feasible. 1653222 11 cfleizach 2020-05-15 16:21:53 -0700 (In reply to Per Arne Vollan from comment #10) > Comment on attachment 399247 [details] > patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=399247&action=review > > R=me. > > >>> Source/WebCore/ChangeLog:3 > >>> + REGRESSION (iOS 13.4.1): SpeechSynthesisUtterance.onend event won't fire on cancel(). > >> > >> Obvious question: can a regression test be made for this issue? > > > > we have one for the Mock synthesizer that goes all through WebCore. This one uses the synthesizer in the client, which AFAICT there were no tests added for. > > > > @Per, @Brent, is there a way to test this flow? Right now we use a mock synthesizer in these tests that bypasses this logic that was fixed > > I am not sure how to write a test for this. Is it possible to write a layout > test that does not use a mock synthesizer? There's some reason we didn't do this, but I can't remember now. Will try to research > > > Source/WebCore/ChangeLog:17 > > + (WebCore::SpeechSynthesis::cancel): > > Perhaps you could include an explanation here if creating a test is not > feasible. Yes 1653236 12 ap 2020-05-15 16:58:54 -0700 Perhaps it could be mocked at a lower level? Trying to actually make sound through device's speakers wouldn't be OK :) Definitely not blocking this fix. 1654802 13 cfleizach 2020-05-20 22:23:14 -0700 (In reply to Alexey Proskuryakov from comment #12) > Perhaps it could be mocked at a lower level? Trying to actually make sound > through device's speakers wouldn't be OK :) > Right, that might be why we moved away from using the actual synthesizer! > Definitely not blocking this fix. 1654804 14 ews-feeder 2020-05-20 22:46:57 -0700 Committed r261984: <https://trac.webkit.org/changeset/261984> All reviewed patches have been landed. Closing bug and clearing flags on attachment 399247. 1654974 15 pvollan 2020-05-21 10:18:02 -0700 (In reply to chris fleizach from comment #13) > (In reply to Alexey Proskuryakov from comment #12) > > Perhaps it could be mocked at a lower level? Trying to actually make sound > > through device's speakers wouldn't be OK :) > > > > Right, that might be why we moved away from using the actual synthesizer! > > > Definitely not blocking this fix. Would it be possible to mute the sound? 1654984 16 cfleizach 2020-05-21 10:36:33 -0700 (In reply to Per Arne Vollan from comment #15) > (In reply to chris fleizach from comment #13) > > (In reply to Alexey Proskuryakov from comment #12) > > > Perhaps it could be mocked at a lower level? Trying to actually make sound > > > through device's speakers wouldn't be OK :) > > > > > > > Right, that might be why we moved away from using the actual synthesizer! > > > > > Definitely not blocking this fix. > > Would it be possible to mute the sound? I think we could set the volume to 0 on all the speech requests 1654994 17 ap 2020-05-21 10:49:38 -0700 Can the synthesizer be invoked multiple times in parallel without flaking? 1654995 18 cfleizach 2020-05-21 10:50:20 -0700 (In reply to Alexey Proskuryakov from comment #17) > Can the synthesizer be invoked multiple times in parallel without flaking? on iOS yes, on macOS I am unsure 399117 2020-05-12 03:44:54 -0700 2020-05-12 03:44:54 -0700 testcase test.html text/html 1057 baba PCFET0NUWVBFIGh0bWw+CjwhLS0gc2F2ZWQgZnJvbSB1cmw9KDAwMzMpaHR0cHM6Ly90YWthby5i YjMuanAvd29yay9zcGVlY2gvIC0tPgo8aHRtbD48aGVhZD48bWV0YSBodHRwLWVxdWl2PSJDb250 ZW50LVR5cGUiIGNvbnRlbnQ9InRleHQvaHRtbDsgY2hhcnNldD13aW5kb3dzLTEyNTIiPjx0aXRs ZT5zcGVlY2hTeW50aGVzaXM8L3RpdGxlPgo8L2hlYWQ+PGJvZHk+CjxidXR0b24gaWQ9InBsYXki PnBsYXk8L2J1dHRvbj4KPGJ1dHRvbiBpZD0ic3RvcCI+c3RvcDwvYnV0dG9uPgo8ZGw+CiAgPGR0 PnNwZWFraW5nPC9kdD4KICA8ZGQgaWQ9InNwZWFraW5nIj5mYWxzZTwvZGQ+CiAgPGR0PnBlbmRp bmc8L2R0PgogIDxkZCBpZD0icGVuZGluZyI+ZmFsc2U8L2RkPgogIDxkdD5wYXVzZWQ8L2R0Pgog IDxkZCBpZD0icGF1c2VkIj5mYWxzZTwvZGQ+CjwvZGw+CjxzY3JpcHQ+CmRvY3VtZW50LmdldEVs ZW1lbnRCeUlkKCdwbGF5JykuYWRkRXZlbnRMaXN0ZW5lcignY2xpY2snLCAoKSA9PiB7CiAgY29u c3QgdXR0ciA9IG5ldyBTcGVlY2hTeW50aGVzaXNVdHRlcmFuY2UoJ0hlbGxvLCB3b3JsZC4gSGVs bG8sIHdvcmxkLiBIZWxsbywgd29ybGQuJyk7CiAgdXR0ci5vbmVuZCA9ICgpID0+IHsKICAgIGFs ZXJ0KCdvbiBlbmQnKTsKICB9OwogIHNwZWVjaFN5bnRoZXNpcy5zcGVhayh1dHRyKTsKfSk7CmRv Y3VtZW50LmdldEVsZW1lbnRCeUlkKCdzdG9wJykuYWRkRXZlbnRMaXN0ZW5lcignY2xpY2snLCAo KSA9PiB7CiAgc3BlZWNoU3ludGhlc2lzLmNhbmNlbCgpOwp9KTsKc2V0SW50ZXJ2YWwoKCkgPT4g ewogIGRvY3VtZW50LmdldEVsZW1lbnRCeUlkKCdzcGVha2luZycpLmlubmVyVGV4dCA9IHNwZWVj aFN5bnRoZXNpcy5zcGVha2luZzsKICBkb2N1bWVudC5nZXRFbGVtZW50QnlJZCgncGVuZGluZycp LmlubmVyVGV4dCA9IHNwZWVjaFN5bnRoZXNpcy5wZW5kaW5nOwogIGRvY3VtZW50LmdldEVsZW1l bnRCeUlkKCdwYXVzZWQnKS5pbm5lclRleHQgPSBzcGVlY2hTeW50aGVzaXMucGF1c2VkOwp9LCAx MDApOwo8L3NjcmlwdD4KCgo8L2JvZHk+PC9odG1sPg== 399215 2020-05-12 18:09:25 -0700 2020-05-12 18:09:25 -0700 testcase2 test2.html text/html 956 baba PCFET0NUWVBFIGh0bWw+CjxodG1sPgo8aGVhZD4KPHRpdGxlPnNwZWVjaFN5bnRoZXNpczwvdGl0 bGU+Cjxib2R5Pgo8YnV0dG9uIGlkPSJwbGF5Ij5wbGF5PC9idXR0b24+CjxidXR0b24gaWQ9InN0 b3AiPnN0b3A8L2J1dHRvbj4KPGRsPgogIDxkdD5zcGVha2luZzwvZHQ+CiAgPGRkIGlkPSJzcGVh a2luZyI+ZmFsc2U8L2RkPgogIDxkdD5wZW5kaW5nPC9kdD4KICA8ZGQgaWQ9InBlbmRpbmciPmZh bHNlPC9kZD4KICA8ZHQ+cGF1c2VkPC9kdD4KICA8ZGQgaWQ9InBhdXNlZCI+ZmFsc2U8L2RkPgo8 L2RsPgo8c2NyaXB0Pgpkb2N1bWVudC5nZXRFbGVtZW50QnlJZCgncGxheScpLmFkZEV2ZW50TGlz dGVuZXIoJ2NsaWNrJywgKCkgPT4gewogIGNvbnN0IHV0dHIgPSBuZXcgU3BlZWNoU3ludGhlc2lz VXR0ZXJhbmNlKCdIZWxsbywgd29ybGQuIEhlbGxvLCB3b3JsZC4gSGVsbG8sIHdvcmxkLicpOwog IHV0dHIub25lbmQgPSAoKSA9PiBhbGVydCgnb24gZW5kJyk7CiAgdXR0ci5vbmVycm9yID0gZSA9 PiBhbGVydChgb24gZXJyb3I6ICR7ZS5lcnJvcn1gKTsKICBzcGVlY2hTeW50aGVzaXMuc3BlYWso dXR0cik7Cn0pOwpkb2N1bWVudC5nZXRFbGVtZW50QnlJZCgnc3RvcCcpLmFkZEV2ZW50TGlzdGVu ZXIoJ2NsaWNrJywgKCkgPT4gewogIHNwZWVjaFN5bnRoZXNpcy5jYW5jZWwoKTsKfSk7CnNldElu dGVydmFsKCgpID0+IHsKICBkb2N1bWVudC5nZXRFbGVtZW50QnlJZCgnc3BlYWtpbmcnKS5pbm5l clRleHQgPSBzcGVlY2hTeW50aGVzaXMuc3BlYWtpbmc7CiAgZG9jdW1lbnQuZ2V0RWxlbWVudEJ5 SWQoJ3BlbmRpbmcnKS5pbm5lclRleHQgPSBzcGVlY2hTeW50aGVzaXMucGVuZGluZzsKICBkb2N1 bWVudC5nZXRFbGVtZW50QnlJZCgncGF1c2VkJykuaW5uZXJUZXh0ID0gc3BlZWNoU3ludGhlc2lz LnBhdXNlZDsKfSwgMTAwKTsKPC9zY3JpcHQ+CjwvYm9keT4KPC9odG1sPgo= 399247 2020-05-13 01:01:21 -0700 2020-05-20 22:46:57 -0700 patch file_211776.txt text/plain 1947 cfleizach ZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZyBiL1NvdXJjZS9XZWJDb3JlL0No YW5nZUxvZw0KaW5kZXggYWYzZGY3ZDI0OTguLjUyYWVkYjY3NTkzIDEwMDY0NA0KLS0tIGEvU291 cmNlL1dlYkNvcmUvQ2hhbmdlTG9nDQorKysgYi9Tb3VyY2UvV2ViQ29yZS9DaGFuZ2VMb2cNCkBA IC0xLDMgKzEsMjEgQEANCisyMDIwLTA1LTEzICBDaHJpcyBGbGVpemFjaCAgPGNmbGVpemFjaEBh cHBsZS5jb20+DQorDQorICAgICAgICBSRUdSRVNTSU9OIChpT1MgMTMuNC4xKTogU3BlZWNoU3lu dGhlc2lzVXR0ZXJhbmNlLm9uZW5kIGV2ZW50IHdvbid0IGZpcmUgb24gY2FuY2VsKCkuDQorICAg ICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9MjExNzc2DQorICAg ICAgICA8cmRhcjovL3Byb2JsZW0vNjMxMzAyNDk+DQorDQorICAgICAgICBSZXZpZXdlZCBieSBO T0JPRFkgKE9PUFMhKS4NCisNCisgICAgICAgIFdpdGggdGhlIG1vdmUgdG8gaGF2aW5nIHNwZWVj aCBzeW50aGVzaXMgaGFwcGVuIGluIHRoZSBjbGllbnQsIHRoZSBjYW5jZWwgY2FzZSBoaXRzIGEg c25hZy4NCisgICAgICAgIFdlIGNhbmNlbCB0aGUgc3BlZWNoIGpvYiBhbmQgY2xlYXIgb3V0IHRo ZSBjdXJyZW50IHV0dGVyYW5jZS4gQnkgdGhlIHRpbWUgdGhlIGNhbmNlbCBjYWxsYmFjayBjb21l cyBiYWNrLA0KKyAgICAgICAgdGhlIGN1cnJlbnQgdXR0ZXJhbmNlIGlzIGdvbmUgYW5kIG5vdGhp bmcgaGFwcGVucy4NCisNCisgICAgICAgIFRoZSBmaXggaXMgdG8gcHJvY2VzcyB0aGUgc3BlZWNo RXJyb3IgZXZlbnQgaW1tZWRpYXRlbHkgYW5kIG5vdCB3YWl0IG9uIHRoZSBzcGVlY2ggc3ludGhl c2l6ZXIgLS0gd2hpY2ggc2VlbXMgc2FuZSwNCisgICAgICAgIHNpbmNlIHdlJ3JlIGp1c3QgY2Fu Y2VsbGluZyBhIHNwZWVjaCBqb2IuDQorDQorICAgICAgICAqIE1vZHVsZXMvc3BlZWNoL1NwZWVj aFN5bnRoZXNpcy5jcHA6DQorICAgICAgICAoV2ViQ29yZTo6U3BlZWNoU3ludGhlc2lzOjpjYW5j ZWwpOg0KKw0KIDIwMjAtMDUtMTIgIEphY29iIFVwaG9mZiAgPGphY29iX3VwaG9mZkBhcHBsZS5j b20+DQogDQogICAgICAgICBVbnJldmlld2VkLCByZXZlcnRpbmcgcjI2MTU1Ny4NCmRpZmYgLS1n aXQgYS9Tb3VyY2UvV2ViQ29yZS9Nb2R1bGVzL3NwZWVjaC9TcGVlY2hTeW50aGVzaXMuY3BwIGIv U291cmNlL1dlYkNvcmUvTW9kdWxlcy9zcGVlY2gvU3BlZWNoU3ludGhlc2lzLmNwcA0KaW5kZXgg NjY4ZDNjYmZkNDEuLjAwYjE4NDMwNTk0IDEwMDY0NA0KLS0tIGEvU291cmNlL1dlYkNvcmUvTW9k dWxlcy9zcGVlY2gvU3BlZWNoU3ludGhlc2lzLmNwcA0KKysrIGIvU291cmNlL1dlYkNvcmUvTW9k dWxlcy9zcGVlY2gvU3BlZWNoU3ludGhlc2lzLmNwcA0KQEAgLTE1MSw2ICsxNTEsOSBAQCB2b2lk IFNwZWVjaFN5bnRoZXNpczo6Y2FuY2VsKCkNCiAgICAgbV91dHRlcmFuY2VRdWV1ZS5jbGVhcigp Ow0KICAgICBpZiAobV9zcGVlY2hTeW50aGVzaXNDbGllbnQpIHsNCiAgICAgICAgIG1fc3BlZWNo U3ludGhlc2lzQ2xpZW50LT5jYW5jZWwoKTsNCisgICAgICAgIC8vIElmIHdlIHdhaXQgZm9yIGNh bmNlbCB0byBjYWxsYmFjayBzcGVha2luZ0Vycm9yT2NjdXJyZWQsIHRoZW4gbV9jdXJyZW50U3Bl ZWNoVXR0ZXJhbmNlIHdpbGwgYmUgbnVsbA0KKyAgICAgICAgLy8gYW5kIHRoZSBldmVudCB3b24n dCBiZSBwcm9jZXNzZWQuIEluc3RlYWQgd2UgcHJvY2VzcyB0aGUgZXJyb3IgaW1tZWRpYXRlbHku DQorICAgICAgICBzcGVha2luZ0Vycm9yT2NjdXJyZWQoKTsNCiAgICAgICAgIG1fY3VycmVudFNw ZWVjaFV0dGVyYW5jZSA9IG51bGxwdHI7DQogICAgIH0gZWxzZSBpZiAobV9wbGF0Zm9ybVNwZWVj aFN5bnRoZXNpemVyKSB7DQogICAgICAgICBtX3BsYXRmb3JtU3BlZWNoU3ludGhlc2l6ZXItPmNh bmNlbCgpOw0K